How third-party AI models can be plug-and-play easily in Scilligence informatics platform. Scientists can leverage embedded AI models to design better drug molecules, minimize wet lab investment, and shorten drug discovery timeline.
A new AI model only helps if it works on the data you already have. In Part 2 of our Fall Webinar Series, Scilligence CEO and co-founder Dr. Jinbo Lee showed how third-party AI models plug into the Scilligence platform, so scientists can design better molecules, reduce wet lab investment and shorten discovery timelines.
To get access to the replay, please fill out the form below.
The recording is available with Chinese, Japanese and Korean Subtitles, please select the replay that works best for you.
What “plug-and-play” means in practice
Scilligence isn’t tied to one model. Teams can connect their own models, or models from specialist partners, through a RESTful API. The models run alongside the other Scilligence applications inside the customer’s firewall, so nothing leaves the environment.
Three examples shown in the webinar
1. Protein structure in RegMol. Pick a protein sequence, choose a prediction model (OpenFold in the demo, NVIDIA BioNeMo as the integration example) and get a 3D structure you can rotate and annotate. The same sequence can run a multiple-sequence alignment against sequences registered in RegMol. AlphaFold and other models can plug in the same way.
2. Property prediction in FOCUS with LexBio.ai. Predictions such as logP, hERG, Caco-2 and CYP inhibition fill the data table beside the calculated properties, so a chemical series can be sorted and ranked. FOCUS runs on the web or as a desktop app for working offline.
3. Retrosynthesis in the ELN with Chemical.AI. From a target molecule, proposed synthetic routes from the ChemAIRS model come straight into the notebook. In the demo, a seven-step route became seven reaction steps, with conditions and linked publications, and colleagues can trace where the idea came from.
Fewer experiments, and results that stay connected
Predicting properties early means fewer experiments on molecules unlikely to succeed. Dr. Lee cited a LexBio.ai example that reached nanomolar inhibitors from about 20 molecules without a large screen, with the caveat that it depends on using the right model correctly. Designs can be registered in RegMol as virtual molecules; when one is made, the physical sample links to the same ID, so designs, results and SAR stay connected.
How easy is it to add a model?
For models Scilligence has already integrated, adding a property is a checkbox and needs no training. Connecting a new model uses the RESTful API and some IT knowledge, and the Scilligence team can help.
Watch the Part 1 replay and register for Part 3
Missed Part 1, AI-Powered Informatics Platform for All Drug Modalities?
The replay is available here: AI Powered Informatics Platform for all Drug Modalities
Part 3, Lab Automation of the Future – When Informatics Meets Robotics, is on Tuesday, November 3, with our CEO Dr. Jinbo Lee and guest speaker Kepin Li, Founder of Hinton.AI.
Register now to save your spot:
2026 Scilligence Fall Webinar Series: Scilligence’s AI-enabled and robotics-integrated informatics platform.
If you’d like to see how a model could connect to your own data, let’s get started:
Book a DEMO Here
Want to see it in action? Fill out the form below to get access to the full webinar replay instantly.
Replay is also available with:
Thank you,
Scilligence Team
